Sign In

Majid Mallis

Lead Software Engineer with: − 15 years of solid experience in Software Development, especially using Javascript and Java. − Infinite learner with an advanced Mathematics background. − Experience in building Software systems at scale. − Good Project/Product Management knowledge. − Mentor and key team player with good communication skills.

Highlights

Jun 27 · linkedin

Reading ML papers these days feels like 😆

P.S. credit for the image is due to its creator.

Mar 16 · linkedin

🚨🚨🚨 Personal announcement 🚨🚨🚨

TL;DR: I am now a "Google dropout" looking forward my next challenge in the Machine Learning space.

Two weeks ago, I announced my resignation from my position at Google and today was my last day at work. The last two years have gone so fast and it was a humbling experience for me. I met lot of smarter and more interesting people, I learned even more about how to build reliable software at scale (in the fintech space), and I had the chance to red-team many AI products before they go live. If I have to redo it over again, there is no question, I will.

The current cycle of AI seems very promising. For the first time it looks like the tech community has cracked the code: the trifecta of data, compute, and algorithms. The increasing interest in AI is stunning and in particular the advances in GenAI are very promising. This isn't another AI cycle, not an iPhone moment, nor an internet moment, not even a printing press moment. This is the steam engine moment! I estimated getting 30% more productive since I started using AI-driven code autocompletion and I'm not alone. The ability to train models with such a high accuracy to the point to disrupt "white collar" jobs presents a serious society shift. More importantly, because it's software, once you train a model, you can spin up as many digital agent instances as you want; if there enough compute power available to your model.

The recent breakthroughs in NLP are just the beginning of a long path towards AGI. Sorry. This is happening. Get over it. And In order to get there, we might be able to create more compute (Moore's law is laughing at me right now) but we definitely need way more scalable algorithms than what we have today. Another problem is hallucination, which is to blame for things like lack of reliability and, even worse, ethical issues. The unprecedented opportunity in front of us, combined with the fact that the field is nascent, made my decision to focus full-time on ML a no-brainer. More on that in a later post.

Now, my focus is on spending more time with my family, getting back in shape, and doing tons of reading and hacking around. If you are learning or working in the ML research space let's chat!

#ai #ml #machinelearning #science #googler

Related Questions

What is Majid Mallis's current role?
How many years of software development experience does he have?
Which programming languages is he especially experienced with?
Does he have experience building software systems at scale?
Majid Mallis
Add to my network

Location

San Francisco, California, United States